摘要
为解决图像配准问题,结合点集GMM建模思想和相干点漂移思想的影响,设计了一种新的点集配准目标函数。首先对特征点集进行归一化处理,然后利用似然函数梯度下降求解配准参数,同时剔除不满足匹配要求的溢出点,利用实际图像序列对算法进行了验证,结果表明该方法具有良好的性能。
For settling the problem of image registration, a new points registration objective function was de- signed with effects of the GMM modeling principles and coherent point drift rules. Firstly, normalizing the feature points, registration parameters are solved by gradient descending of likelihood function and then the overflow points which cannot meet the requirements the registration are removed. The arithmetic is verified with the actual image sequence that results show good performance.
出处
《科学技术与工程》
北大核心
2013年第13期3768-3773,共6页
Science Technology and Engineering
关键词
运动平台估计
背景补偿
图像配准
相干点漂移
point drift motion platform estimate background compensating image registration coherent
